A note on Malliavin fractional smoothness for Lévy processes and approximation (Q372808): Difference between revisions

The authors consider a Lévy process \(X = (X_t)_{t \in [0, 1]}\) that is an \(L^2\)-martingale and \(Y\) as the stochastic exponent of \(X\) or \(X\) itself. For certain integrands \(\varphi = (\varphi_t)_{t \in [0, 1]}\), the authors investigate the behaviour of \( \left\|\int_{(0, 1]} \varphi_t \, dX_t - \sum_{k=1}^N v_{k-1} \, (Y_{t_k} - Y_{t_{k-1}}) \right\|_{L^2}, \) where \(v_{k-1}\) is an \({\mathcal F}_{t_{k-1}}\)-measurable random variable, in dependence on the fractional smoothness in the Malliavin sense of \(\int_{(0, 1]} \varphi_t dX_t\). Such techniques appear when the stochastic integral is obtained by the Galtchouk-Kunita-Watanabe decomposition of a random variable \(f(X_1)\). Using the example \(f(X_1) = I(K < X_1 < \infty)\), it is shown how fractional smoothness depends on the distribution of the Lévy processes.